Something shifted for me recently. This week, for the first time, I could actually see myself walking away from the flagship corporate AI products. Not from AI itself — from the companies.

I’ve been a genuine enthusiast for AI tools for a few years now. I’ve run training sessions for colleagues, brought it into my classroom, and helped nonprofit fundraisers see what these tools might do for their work. I’ve believed, and still believe, that there’s something real and valuable here. So, how on earth did I get here?

The school year is ending, and when I looked back at it, I remembered that last year I collaborated with five or six teachers on AI projects. This year, one. Early in the year I ran a training that got a lot of pushback (some of it surprisingly strident). There were (mostly younger) teachers who were resistant to what the technology represents: something being imposed rather than offered. At the time I thought I understood. Now I think I actually do.

We are not being invited to consider AI. We are being told it’s here and we have to deal with it.

That’s not just coming from school administrators or policy documents. It’s coming from the companies themselves. The way OpenAI, Google, Anthropic, and the others talk about what they’re building, it’s often addressed to an abstraction called Humanity, not to actual people with actual concerns. There’s a piece by Jim Prosser that frames this through the lens of Dr. Manhattan from Watchmen: the nuclear physicist who gains godlike power and gradually loses the ability to care about individual people. The view from orbit is just too intoxicating. He can see the arc of civilization. He just can’t sit with someone who’s grieving.

That’s a disturbingly accurate portrait of how the leaders of these companies communicate. They care enormously about Humanity. They’d do anything for Humanity. They just can’t be bothered with actual people who are worried about their jobs, their kids’ learning, and their creative work being scraped without consent.

Prosser draws a direct parallel to the American nuclear industry, which ran exactly this kind of communications strategy, talking over the public’s head about civilizational transformation, never building any reservoir of genuine goodwill with ordinary people, and then watched the whole thing collapse when accidents confirmed everyone’s worst fears. The communications failures didn’t kill nuclear power; the disasters did. Decades of dismissing public concern meant there was nothing to draw on when things went wrong.

I can see that scenario unfolding with AI. A catastrophic security incident. The investment bubble bursting, with taxpayers absorbing the losses while the billionaires walk away. Either one could poison the whole thing, the companies and the technology alike, including the parts that are genuinely remarkable.

By using these products, am I part of the problem?

I’m not there yet. At least I don’t think I am, not quite. The environmental impact of my personal use is negligible compared to the structural decisions made long before I opened a chat window. We live inside systems we didn’t choose. We never decided on single-use plastics either, and I’m skeptical of the personal-responsibility framing that puts the weight of industrial choices on individual consumers. Score-keeping gets stupid fast.

The contradiction is real, though. I had a session a few weeks ago helping nonprofit fundraisers think through how AI tools might help them, as I mentioned in my last post. These are organizations always stretched too thin, doing work that genuinely matters. I walked away feeling like I’d put something good into the world. Now I’m not entirely sure that’s true.

The technology itself is relatively neutral. It can be used to make the world better or worse. The creation of it was not neutral, however: the training data, the decisions about reinforcement learning, and the non-consensual scraping underlying all of it make it decidedly opinionated in ways that threaten to increase inequality. The technology is certainly extraordinary in what it is. It’s fairly ordinary in how it can be used, though, meaning a person with good judgment and honest intentions can use it honestly. The companies that built it are making choices that deserve scrutiny.

So I’m building a new course.

It’s going to be called Thinking Through AI, which is also what I’m doing right now, and that feels right. I want to create the space I haven’t been able to find in conversations with colleagues or in the broader culture, where students can actually work through this together, use the tools, examine the impact on society, and draw their own conclusions from actual experience and knowledge rather than vibes.

They have more at stake in this than I do — I’m near the end of my career with all of this emerging around me, and they’re at the beginning. I genuinely want to hear what they think. If they decide at the end of the course that AI isn’t for them, that’s a completely valid outcome, as long as it comes from understanding and experience rather than reflexive dismissal.

Something unexpected gave me a bit of hope this week. The Pope released an encyclical on AI, and I found myself reading it with a lot of recognition. It centers on using the technology to make the world better and warns explicitly against AI becoming another engine of rising inequality, helping the already-advantaged and leaving everyone else behind. What struck me was the source: not a corporation, not a political party, and not a tech pundit. One of Anthropic’s co-founders was present for the release, which tells me there are at least some people in this industry who aren’t fully captivated by the view from orbit.

If things don’t change, what walking away actually looks like for me isn’t walking away from AI. It’s walking away from the specific companies. I’ll try local models on my own hardware, or open-weight models through API access. The work will continue either way, without feeding the investment bubble or the toxic billionaires running it.

I’m not there yet. But I can see it now in a way I couldn’t before.